PDA

View Full Version : exporting video seq for planer texture



blindsided
04-08-2009, 12:41 PM
this is a repeat from another thread that i think may have died a long time ago, just thought i would post again as i would love some opinions.




I am making a 3d book, roughly A4 shape.

On the pages I am going to have video playing in a frame, surrounded by eye candy stills and text.

So the animation roughly goes.... book opens, we flip through a couple of pages, finish on the main page, camera slowly tracks into to a box playing video.

The way I created it is by making an AFTER EFFECTS comp the aspect ratio of my page.
PLACED MY VIDEO IN POSITION ON THE PAGE
RENDER OUT AS A SERIES OF FRAMES
APPLIED THE SEQUENCE AS AN IMAGE MAP TO THE PAGE

My problem is field rendering!!!!

The video is 8 BIT PAL 16/9FH UPPER FIELD FIRST from Final Cut Pro
and needs to be placed back in the 8 Bit sequence after rendering.

should I render the image seq from AE UPPER FIELD?
should I render my LW scene UPPER FIELD?

My initial tests were render the image seq containing the video as upper from AE, then interpret the seq in IMAGE EDITOR as upper, then render out as upper ( as a TGA image seq) from LW, back into AE to make movie and into FCP..maintaining UPPER all the way through.

The result however is not great. The video looks very soft and fuzzy.
Am trying to render out the inital page seq as no fields, but not sure this will just dump 1 of the video fields, and I will end up with strobing?

This has been a problem I have encountered several times, and always feel I am compromiseing the outcome rather than maintaing a solid workflow.

any help or advice would be most welcome.
blindsided is online now

Sarford
04-10-2009, 10:41 AM
I think you should de-interlace your footage. Shake has some very decent de-interlace routines which don't just dump even or uneven fields.
Interlaced footage inside 3D is always a problem couse it projects the maps as progressive footage, meaning you get jitters on the sides (the overlapping fields), which can give funny results when the image isn't 1:1. If you only have AE, I'm sure there must be some decent de-interlace plug-ins for that as well.

This whole interlace crap should come to an end soon, I hate it!

JeffrySG
04-10-2009, 11:04 AM
Yeah, I totally agree with de-interlacing your footage that goes on the page. As that footage will be moving, rotating and scaling on the screen as you zoom into it, the fields will not even be real fields on the final output making it look even more like crap after it's rendered. Like Sarford said, if that footage isn't exactly 1:1 with the final output it would never look correct.

erikals
04-10-2009, 11:26 AM
don't use fields, that's an old technique :)

blindsided
04-10-2009, 02:41 PM
At present the best technique I have found is...

INJEST USING FCP BLACKMAGIC PAL FH ANAMORPHIC 8BIT CODEC 720x576 UPPER FIELD

IMPORT INTO AE INTERPRET FOOTAGE CORRECTLY

CREATE AN A4 SIZE COMPOSITION SQUARE PIXELS

COMPOSITE VIDEO AND GRAPHIC ELEMENTS IN COMPOSITION

RENDER OUT TARGA SEQ NO FIELDS

IMPORT SEQ INTO LW APPLY SEQ AS TEXTURE TO PAGE

RENDER OUT USING PAL D1/DV WIDESCREEN DEFAULT NO FIELDS PERSPECTIVE CAM, AS 0.02 OS 0.02 PHOTOREAL MOTION BLUR TARGA SEQ

IMPORT INTO AEP INTERPRET CORRECTLY RENDER OUR UNCOMPRESSED QT NO FIELDS

I will keep trying alternates to see if I can get any better results, the final footage in FCP has a certain 'look and feel' similar to the results of the video to film look you get with MAGIC BULLET et all. So although its not horrible interlace strobing, it does feel a little bit of a work around.

thanks all for the advice i will post up the results after the easter break if u r interested.